A math book usually represents a way of thinking about a topic, a perspective on that topic. Hence, you have to agree with the authors opinion on how to present the topic in order to get a useful learning approach. (Contrary, if you completely disagree with the book's perspective, you may regard it as a challenge.)
In my experience, learning achievements are enhanced if you let the knowledge flow through you own hands. This means you have to put down the content of the book in a way which fits your way of thinking best. The author will probably have a (slightly) different perspective than you have, due to taste and ability.
A good approach towards a book is: "The author is lying." - each line of the book has to be justified. If you can't do so, you do not understand the topic in full.
Furthermore, I have made the experience it is inevitable to spend lots of time with the matter. Difficult and inaccessible proofs may unveil if you read it over and over again (imo, good examples of these are Hörmander's books on linear pde).